Abstract
The public sphere often struggles with a trade-off between authentic, small-scale discourse and broad, superficial debate dominated by influential entities. This study explores how media and information literacy (MIL) can foster a public sphere that is both expansive and genuinely representative. I investigate this by using Large Language Models (LLMs) to analyze a vast body of scholarly literature and develop a model explaining MIL’s positive impact. Employing a qualitative grounded theory methodology, I used ChatGPT, DeepSeek, and Gemini to perform open, axial, and selective coding on 7,670 abstracts from the Web of Science database published within the last five years. While the LLMs’ coding frameworks differed slightly, they demonstrated significant agreement on abstracts concerning MIL’s role in expanding public discourse. The analysis yielded five to seven core thematic categories and numerous subcategories. Notably, both ChatGPT and Gemini converged on similar overarching frameworks, identified as “civic resilience in the digital age” and “empowered media citizenship.” These concepts align with existing literature linking MIL to human rights, digital citizenship, and sustainable development. The study suggests that a multi-stage grounded theory approach, analyzing smaller literature batches with sophisticated prompt engineering, could yield even deeper insights as LLM technology advances.
1. Introduction
The public sphere can manifest in two distinct, yet problematic, forms. It may be characterized as authentic yet constrained, or conversely, as expansive yet illusory. In the former scenario, a genuine space for freedom and public discourse exists, but its influence remains limited and it does not constitute a truly open public sphere. In the latter, the public sphere presents an appearance of broad societal debate on diverse issues. However, this appearance is deceptive, as the discourse is monopolized by a restricted set of forces, groups, institutions, and influential figures. This raises a critical question: What mechanisms can ensure the public sphere is simultaneously broad in its reach and authentic in its expression, freely and clearly reflecting public opinion without being subject to monopolization or manipulation?
Jürgen Habermas (1964) conceptualized the public sphere as “a realm of our social life in which something approaching public opinion can be formed,” where “access is guaranteed to all citizens” and which emerges “in every conversation in which private individuals assemble to form a public body” (p. 49). He further cautioned that the concentration of media control and the power exerted through it could precipitate a “refeudalization” of the public sphere (Habermas, 1964, p. 54).
Jürgen Habermas’s seminal work on the public sphere, first articulated in the 1960s, has become a cornerstone of inquiry for scholars across numerous disciplines, including political science, economics, media studies, and sociology. The contemporary relevance of this research, particularly from a media and communications perspective, has been significantly amplified by the recent global information and technological revolution. This includes advancements in the communications sector, as well as the rapid development of artificial intelligence and internet technologies.
Analyzing the transformation of the media’s role and its subsequent impact on the public sphere, Jologua (2019, pp. 107-109) distills Habermas’s critical theses. The analysis posits that the media have undergone a fundamental shift, becoming instruments driven by commercial and political imperatives, which now address the public in a manner analogous to how advertisers target consumers.
This transformation has led to a degradation of the media’s capacity to facilitate rational-critical discourse among citizens. Consequently, the public has been relegated to the status of a passive recipient of information rather than an active participant in debate. The media, in this view, have evolved into a mechanism for political and economic elites to manufacture public consent. This has fostered an environment of manipulation, thereby diminishing the media’s critical function.
Historical and cross-societal analyses reveal diverse and often conflicting trends in the relationship between media and the public sphere. From a media literacy standpoint, the Council of Europe’s integration of media literacy into its public policies exemplifies a positive intervention, fostering a more resilient public sphere by equipping citizens to combat disinformation. In response to the pervasive issue of disinformation, policymakers within the European Union have enacted various measures, including the regulation of online platforms, the advancement of media literacy initiatives, the combating of disinformation, and the endorsement of fact-checking organizations. Nevertheless, the successful implementation of policies to counter disinformation remains limited to a small number of nations (Bleyer-Simon et al., 2024; O’Neill & Barnes, 2007).
Nagaraj et al. (2014) assert that within the Indian public sphere, media literacy is significantly marginalized. They argue that this sphere is contracting due to the ascendancy of private media ownership and prevailing market forces. Furthermore, they characterize the public sphere as a contested space marked by conflict between corporate endeavors and civil society. Building on this, Jayachandran (2018) analyzes media literacy in India through a multidimensional lens, considering its interplay with the public sphere, market dynamics, technological advancements, access disparities, civil society engagement, and citizen activism. He proposed comprehensive, multifaceted, and sustained approaches for cultivating critical awareness.
The contemporary information environment has been marked by an unprecedented proliferation of misinformation, disinformation, and what is known as “information disorder” Media and information literacy is therefore a critical competency, enhancing individuals’ capacity to critically evaluate and counter deceptive content. Moreover, it strengthens public oversight of institutions and expands the circle of influencers in social media. A media-literate generation, empowered by the global ubiquity of mobile devices and expanding internet connectivity, possesses the potential to fundamentally reshape the global public sphere (Mansoor, 2024; Mihailidis, 2015).
The extent to which enhancing media literacy contributes to a more robust public sphere remains a subject of scholarly inquiry. A critical analysis of the extant literature is necessary to determine if a comprehensive theoretical model can be developed to guide future research and inform policy.
2. Theoretical Background
2.1. Public Sphere and Media Literacy
The concept of the public sphere was systematically developed by the German philosopher and social theorist Jürgen Habermas in his seminal 1962 work, later translated into English in Habermas, 1989 as “The Structural Transformation of the Public Sphere: An Inquiry into a Category of Bourgeois Society”. As a prominent second-generation member of the Frankfurt School, Habermas has formulated critical social theses that have had a profound interdisciplinary impact, influencing fields ranging from communication and cultural studies to political science and democratic theory.
His intellectual trajectory is marked by a departure from foundational Marxist tenets, which facilitated the development of a distinct critical theory that engages seriously with liberal thought, a trend particularly evident since the late 1980s (Jologua, 2019; Sharlamanov, 2024).
Habermas’s critical perspective on the media’s detrimental influence on the public sphere finds its intellectual lineage in the work of Theodor Adorno, a foundational figure of the Frankfurt School and a trenchant critic of mass culture (Lubenow, 2012). This critical stance on the media’s role in the politicization of the public sphere has remained a consistent element of Habermas’s work for half a century, from 1962 to 2012. Although Habermas’s later Theory of Communicative Action appears to diminish the centrality of media ( Habermas, 1984 ), scholars such as Douglas Kellner have suggested that the mass media in particular serve “as the major instrument through which Habermas’s system invades and dominates the lifeworld” (Jologua, 2019, p. 112).
Following the English translation of Jürgen Habermas’s The Structural Transformation of the Public Sphere, scholarly critiques of his model promptly ensued. These criticisms, as noted by Budarick (2016), primarily contested the normative nature of the “public,” the model’s insufficient theorization of power, its privileging of rational-critical debate, and the feasibility of achieving social consensus in pluralistic societies. Moreover, the foundational bourgeois-liberal assumptions of Habermas’s conception were a central point of contention.
The Habermasian concept of a national public sphere, traditionally forged by print and broadcast media, has undergone a significant transformation with the rise of social media into a multi-layered sphere. As Iosifidis (2011) contends, the anticipated democratizing and empowering effects of the internet and social media are often overstated. Unfettered participation can lead to anarchic conditions, and the digital realm has increasingly become dominated by corporate and political interests. Consequently, the extensive dialogue and critical debate that are foundational to the public sphere are frequently diminished in online environments. Both Fuchs (2020) and Iosifidis (2011) posit that the cultivation of a genuinely inclusive public sphere is contingent upon the contributions of robust public service media and critical alternative media outlets.
The conception of media literacy has undergone a significant evolution, adapting to transformations within communication and cultural milieus. Initially framed as a defense mechanism against harmful communication messages, the concept has broadened to encompass the proficient creation of critical and analytical messages. As such, Media and Information Literacy (MIL) emerges as a crucial framework for augmenting the capacity of individuals and communities to navigate and counteract communication disorders. MIL enhances societal resilience in three principal domains: fostering resistance to disinformation, cultivating the skills necessary for critical examination and oversight of dominant societal institutions, and mitigating the effects of monopolistic control within social media platforms (Mansoor, 2024).
Habermas (2022) articulates a central paradox: digitalization simultaneously broadens the scope of the public sphere while inducing its fragmentation, a process that universally enfranchises individuals as communicators. In her response, Cristina Lafont (2023) concurs with Habermas on the necessity of an evolutionary development of media competencies among digital actors. She posits, however, that the causal factors of public sphere fragmentation and polarization may render interventions such as media literacy education or platform governance inadequate to counter the escalating trend of citizen self-isolation across both online and offline contexts. Yet, as Han (2024) observes, the very media literacy that Habermas (2022) indirectly suggests constitutes both “a hope and a condition for citizens to become subjects of democracy despite the turbulent environment”.
2.2. Public Sphere Expansion
Contemporary structural transformations of the public sphere are precipitated by three primary developments: digitalization, commodification, and globalization. Seeliger and Sevignani (2022) argue that the bidirectional nature of digital media facilitates mass access, thereby expanding the public sphere. This expansion blurs the traditional demarcation between individual, group, and mass media discourses—particularly regarding the intersection of private and public domains—resulting in the emergence of a plurality of public spheres. Historically, Habermas identified the expansion of the welfare state, economic commercialization, and media evolution as pivotal factors influencing the public sphere (Sharlamanov, 2024). However, a central concern for Habermas was the potential trade-off between scale and quality; specifically, he questioned whether the expansion of the public sphere has deleteriously affected its capacity for rational-critical discourse (Calhoun, 2017).
McNair et al. (2017) propose an evolutionary model of the public sphere categorized by the dominant medium of the era: • The National Public Sphere: Defined by the dominance of print media. • The Transnational Public Sphere: Facilitated by the rise of satellite television. • The Globalized Public Sphere: Emergent with online news platforms. • The Networked Public Sphere: Characterized by the age of social media.
However, scholarly consensus regarding the networked sphere remains elusive. While some research suggests that social networks mitigate geographical barriers to foster a truly transnational sphere, other findings indicate a persistent adherence to national and local online spheres (Samuel-Azran & Hayat, 2020).
The expansion and contraction of the public sphere demonstrates its elasticity. This dynamic nature is evidenced by its elasticity regarding ethnic minorities, transnational media, and diaspora communities. Driven by expatriate actors and cross-border media, the public sphere transcends traditional boundaries to encompass transnational and global dimensions. Budarick (2016) delineates two primary shifts in the conceptualization of the public sphere. The first pertains to the fragmentation of the unitary sphere into discrete sub-spheres, a phenomenon driven by sub-national identities and the proliferation of ethnic media. The second involves the transcendence of national boundaries, signaling the emergence of a transnational public sphere.
As the media landscape grows increasingly complex, the debate surrounding the expansion, contraction, and societal implications of the public sphere remains a vital area of inquiry.
3. Previous Studies
Jürgen Habermas’s foundational conceptualization of the public sphere has precipitated extensive cross-disciplinary discourse. Hofmann’s (2021) bibliometric investigation into the reception of Habermas within the United States (1974–2018) substantiates the enduring prominence of The Structural Transformation of the Public Sphere, with 22,000 citations recorded by April 2018.
In his 2004 work, Lincoln Dahlberg delineated a comprehensive framework of the Habermasian public sphere, intended for the critical assessment of contemporary communication practices. Dahlberg articulated a set of normative conditions that define the ideal for discourse within this sphere. These stipulations include: the rigorous and reasoned critique of questionable validity claims; a reflexive process wherein participants critically evaluate their own values and standpoints; the adoption of an ideal role-taking perspective, which obligates participants to earnestly strive for a comprehensive understanding of opposing viewpoints; an unwavering commitment to sincerity, precluding all forms of interpersonal deception and self-deception; the assurance of both formal and discursive equality, guaranteeing the equitable participation of all germane parties; and the preservation of autonomy from governmental and corporate influence, ensuring that the discourse remains independent of the state and economic power structures.
Lorenzano et al. (2018) investigated the influence of social media platform affordances, such as interactivity and engagement, on the online public sphere during the 2016 U.S. presidential election. Set against a backdrop of increasing political polarization and online apathy, the study employed in-depth interviews with politically engaged youth. Through a deductive coding analysis of the interview transcripts, the findings indicated that participant experiences largely failed to satisfy the normative criteria for the public sphere as articulated by Dahlberg (2004).
The imperative of equitable representation constitutes a critical dimension within the field of media literacy, particularly in relation to the public sphere. Research focusing on this dimension includes studies by King and Mele (1999) and Caton-Rosser (2006), which investigated community media and the active engagement of its users in socio-politically driven power dynamics. This focus is pertinent given the corporate concentration dominating mainstream 21st-century media. These studies concluded that community media, specifically community television, offers a platform for ordinary citizens and marginalized voices otherwise excluded from mainstream discourse. Consequently, these findings necessitated a redefinition of the traditional public sphere to encompass “purposeful action by local citizens from diverse backgrounds.”
Benson’s (2008) study critically engaged with Habermas’s perspective, which posits that escalating commercialization compromises the media’s capacity to function as a public sphere—that is, to facilitate rational, critical public debate with maximal citizen participation. The study employed a comparative methodology, testing hypotheses concerning commercial and state influences on news production by contrasting the commercially driven American media system with the state-dominated French system. The analysis of news coverage regarding protest events revealed that French media exhibited greater participatory qualities, were less rational in specific aspects, yet maintained an equivalent level of criticality.
According to Benson (2008), the French media system, being less influenced by commercial interests, was notably more participatory: it incorporated a broader spectrum of institutional and ideological viewpoints and demonstrably fostered higher levels of political engagement. However, its degree of rationality was not significantly superior. While it demonstrated procedural rationality in its utilization of narrative techniques, its quotational practices mirrored those of the American press, arguably rendering its dramatic presentation of political conflict less rational. Benson (2008) sought to remedy what was perceived as Habermas’s neglect of “national or other cultural particularities” in his public sphere theory, emphasizing how transnational variations in media systems either facilitate or impede participatory, rational, and critical public political discourse.
Concerning the positive contributions of both traditional and digital media to the public sphere, prior research has underscored the role of satellite television in the mid-1990s in fostering a novel public sphere. Furthermore, the advent of the online public sphere has enabled citizens to bypass and even contest established controls (Douai & Nofal, 2012). Social media platforms also play a significant role in establishing a prominent public sphere, even within authoritarian contexts (Nangi et al., 2024; Samuel-Azran & Hayat, 2020). Nevertheless, this critical function is subject to various constraints and challenges, including the proliferation of disinformation, misinformation, opinion polarization, and conflicts of interest.
Ultimately, these opportunities, constraints, and challenges constitute the core concerns of media literacy, through which the public sphere can be effectively served in a manner consistent with the promotion of the public good.
Nasiri et al. (2024) investigated an optimal framework for media literacy to support the development of a virtual public sphere during the COVID-19 crisis. Their research posits that media literacy is essential for participation in a networked society and a critical component of digital citizenship, thereby facilitating the creation of a virtual public sphere. The study concludes that the pandemic-era virtual public sphere was deficient, primarily due to a lack of applied media literacy education and the high prevalence of disinformation. The authors recommend a tripartite model for improving media literacy, targeting individuals, officials, and the media itself.
While social media platforms present opportunities for fostering rational and critical discourse in a manner analogous to the public sphere, their efficacy is impeded by the proliferation of misinformation, ideological polarization, and the concomitant rise of extremism and hate speech. In response to these challenges, Kumar and Maurya (2024) advocate for a multi-faceted approach. Their proposed framework encompasses the reinforcement of professional journalism, the implementation of rigorous fact-checking mechanisms, the deployment of platform-specific and technological interventions, the enforcement of pertinent legal and regulatory frameworks, and the cultivation of media literacy.
An examination of the interplay between media literacy and the Habermasian concept of the public sphere reveals a significant potential for mutual reinforcement. Jologua’s (2019) thesis investigates this interrelationship, positing that media literacy can serve as a crucial instrument for the enhancement and revitalization of the public sphere in contemporary society. The study demonstrates that the core tenets of media literacy education align with and can effectively address the complex issues articulated by Jürgen Habermas concerning the media’s role in the public sphere. By fostering a critical understanding of media systems and messages, media literacy education can heighten public awareness of the forces that shape public discourse, thereby contributing to the reorganization and strengthening of the public sphere. Jologua’s research concludes that the integration of media education competencies across all societal levels—from primary schooling to university and adult education—is paramount for realizing this potential. The study delineates a dual-faceted approach to media literacy in this context. Firstly, a protectionist framework for media literacy is proposed as a mechanism to counteract the strategic actions employed by political, economic, and media systems. These actions are often aimed at consolidating dominance and monopolizing influence over public opinion. Secondly, a participatory model of media literacy is advocated to foster communicative action within public life. This approach prioritizes the development of knowledge, interactive dialogue, and reasoned debate, which are the foundational elements of a robust and functioning public sphere as envisioned by Habermas.
Previous scholarly work has consistently highlighted the significance of media literacy in fostering a robust public sphere that upholds the public interest. However, this focus has evolved beyond media-specific skills to encompass a broader framework of literacies, serving as adaptive responses to changing needs within the public sphere. This expanded perspective incorporates concepts such as information literacy (Andersen, 2006) and data literacy (Markham, 2020). Critically, these approaches emphasize that effective engagement is not solely predicated on technical research skills. Instead, it necessitates a deep understanding of the associated social and political competencies. This trend aligns with developments in the media literacy literature itself, which is increasingly situated under the more comprehensive umbrella term of media and information literacy (MIL).
4. Research Gap
While contemporary media have ostensibly expanded the public sphere, their role remains a subject of critical debate. Persistent concerns regarding inequitable access, skewed representation, and the erosion of rational-critical discourse on public matters are frequently attributed to media involvement (Oxford Reference, 2024). This expansion is also linked to the formation of what some scholars term “oppositional public spheres” (Livingstone & Lunt, 1994). According to this perspective, the classic Habermasian bourgeois public sphere, by its nature, tends to marginalize or exclude dissenting voices. Consequently, these excluded groups may establish alternative forums for discussion. These alternative spaces emerge as oppositional public spheres and thus expanding the existing public sphere.
According to Enjolras and Steen-Johnsen (2017), the public sphere’s deliberative capacity is particularly challenged by “the problems of fragmentation and representative democracy” The digitalization of public life has profoundly impacted how ideas are distributed, opinions are shaped, and identities are articulated, which in turn affects civic practices and democratic institutions. The ultimate consequences of this transformation are still ambiguous, holding the potential for both beneficial and detrimental developments.
As the media and communications environment undergoes successive and complex changes, there is a sustained scholarly debate concerning the constitution of the public sphere, the factors contributing to its expansion or contraction, and its overall influence on society.
Extensive scholarship has established media literacy as a critical intervention for the expansion and fortification of the public sphere (O’Neill & Barnes, 2007; Erstad & Amdam, 2013; Mihailidis, 2015; Jologua, 2019; Tomšič, 2023; Bleyer-Simon et al., 2024; Han, 2024; Kumar & Maurya, 2024; Nasiri et al., 2024). However, extant literature has yet to adequately elucidate the precise mechanisms governing this process, leaving the field without a cohesive model that explicates media literacy’s functional role. Even proximate contributions, such as Jologua (2019), have not fully synthesized a comprehensive framework.
The substantial volume of existing research offers a prime opportunity to bridge this gap through the application of Grounded Theory methodology. Recognizing that manual analysis imposes inherent limitations on the scope of literature review, this study integrates artificial intelligence to transcend human cognitive constraints. By employing AI to facilitate rigorous open, axial, and selective coding, the research aims to systematically process an expansive corpus of literature, ensuring the construction of a robust, methodologically sound model.
Despite the prevalence of studies utilizing grounded theory in media literacy (Alighardashi et al., 2021; Baum, 2021; Byeon, 2022; Montazeri et al., 2022; Saravani et al., 2025) and public sphere discourse (Dutta-Bergman, 2005; Rashid & Asif, 2025; Slavtcheva-Petkova, 2023), these scholarly works have yet to explore the potential of artificial intelligence models for conducting large-scale qualitative analysis. A critical gap remains in synthesizing the extensive research of recent decades to construct a theoretical model explaining how improved media literacy facilitates an expanded public sphere.
This study aims to fill this research gap by answering a key question: How can Large Language Models (LLMs) be employed to analyze a vast corpus of scholarly literature to develop a model that explains the mechanisms by which media literacy positively influences the public sphere?
In accordance with a grounded theory methodology, the research question is intentionally broad at the outset of the study. It is then progressively refined and focused as conceptual categories and their relationships emerge from the iterative process of data analysis. Therefore, the present study commenced with a general, open-ended inquiry, aligning with the principle that a research question in qualitative research serves as “a statement that identifies the phenomenon to be studied” (Strauss & Corbin, 1998, p. 41).
5. Methodology
This research employs an LLM-assisted qualitative meta-synthesis informed by grounded theory (GT) principles. While Strauss and Corbin (1990) define GT as a rigorous set of procedures to inductively construct a theoretical model from primary data, establishing a comprehensive explanatory model for complex phenomena—such as the Public Sphere—is rarely achievable within the confines of a single empirical study. Consequently, a qualitative meta-synthesis represents a more pragmatic objective. By leveraging the extensive corpus of existing scientific literature, this approach effectively draws upon the inductive, data-grounded spirit of GT. This approach is further situated within a Critical Realist paradigm. Oliver (2012) posits that Critical Realism and Grounded Theory share a substantial compatibility through their mutual focus on iterative theory refinement and praxis. Applied to a meta-synthesis, this synthesis—often articulated as Critical Realist Grounded Theory—anchors the research in a robust epistemology while directing Critical Realism toward applied qualitative analysis (Roberts, 2014).
Drawing parallels to the established coding stages of traditional GT, Hoddy (2019:115) identifies four corresponding stages within Critical Realist research: 1. Description: The documentation of events or situations grounded in the interpretations and accounts of social actors as recorded in the extant literature. 2. Analytical Resolution: The identification and dissection of the phenomenon’s constituent elements. 3. Abduction and Retroduction: The theoretical redescription of these components to postulate the underlying structures and generative mechanisms governing them. 4. Concretisation and Contextualisation: The examination of how these abstract components and mechanisms interact to produce effects within specific, concrete situations.
To operationalize this qualitative analysis across a vast body of literature, the study adopted a systematic protocol utilizing large language models (LLMs)—specifically ChatGPT, DeepSeek, and Gemini—as assistive analytical tools.
Within this context, the study draws conservative inspiration from the emergence of “computational grounded theory.” As articulated by Nelson (2020) and Tschisgale et al. (2023), integrating computational techniques allows researchers to extract patterns from datasets whose magnitude precludes manual analysis. In this LLM-assisted workflow, the computational models assist in initial pattern recognition and semantic clustering across the literature corpus, while the human researcher maintains interpretive authority.
Fundamentally, this approach does not seek to automate qualitative inquiry, but rather to cautiously embed artificial intelligence within the human analyst’s workflow. While computationally managed initial passes can help surface latent themes and mitigate certain structural constraints, the findings remain strictly provisional. The analyst subsequently evaluates the validity of model-generated categories, deepening interpretive understanding through a close reading of the primary texts. Ultimately, while AI provides advanced data-processing capabilities, synthesizing these patterns into novel theoretical frameworks necessarily relies upon the human researcher’s specialized domain expertise and theoretical sensitivity (Carlsen & Ralund, 2022; Nelson, 2020; Van Veggel et al., 2025; Wen et al., 2026).
Because traditional GT demands profound immersion in the data, AI-enabled coding introduces the risk of overlooking critical contextual nuances. Therefore, this study adopted a hybrid methodological approach: Large Language Models (LLMs) were utilized as heuristic aids during the preliminary stages of open coding and idea generation. Conversely, the advanced stages of conceptual encoding relied more on the researcher’s interpretive and synthetic faculties, ultimately culminating in the construction of the theoretical model and the elucidation of its broader academic and practical implications.
5.1. Data Collection
A systematic and comprehensive inventory was conducted on all research outputs indexed in the Web of Science database under the keyword “media literacy”. The search, executed on May 14, 2025, yielded a total of 7670 documents published between 2020 and 2025. This timeframe was selected because the past six years encompass the majority of media literacy research indexed in the Web of Science. Specifically, 7,670 of the 14,983 identified results (51.2%) were published during this period. Furthermore, this timeframe is particularly relevant for examining public sphere literature given the concurrent transition to digital infrastructure (Seeliger & Sevignani, 2022; Staab & Thiel, 2022). Recent scholarship has increasingly scrutinized the virtual public sphere (Nangi et al., 2024; Nasiri et al., 2024), while Habermas (2022) has highlighted how digitization expands and fragments the public sphere, effectively transforming all participants into active communicators.
Sample Characteristics
5.2. Coding Procedures via LLMs
The data analysis and coding procedure in this study adheres to the GT methodology, which unfolds across three primary stages (Bryant & Charmaz, 2007; LaRossa, 2005; Mansoor, 2024; Strauss & Corbin, 1990, 1998; Corbin & Strauss, 2008): • Open Coding: This initial stage involves the conceptualization of the raw data. The data is broken down to identify, name, and categorize phenomena, and the properties and dimensions of these categories are elucidated. • Axial Coding: In the second stage, the data is reassembled in new ways by making connections between a category and its subcategories. This process occurs around the “axis” of a category, linking them at the level of their properties and dimensions to build a more complex and integrated understanding. • Selective Coding: The final stage requires the identification of a “central category”—the core phenomenon around which all other categories are integrated. This central category must demonstrate “analytical power” to refine and integrate the categories into a cohesive theoretical framework, thereby forming a single explanatory unit.
The analytical framework of this research employs LLMs, departing from traditional qualitative techniques which are rendered impractical by the sheer volume of the dataset (n=7670). Conventional methods are not only resource-intensive but also present challenges related to interpretive consistency and potential bias. In contrast, LLMs, which are underpinned by advanced generative AI, provide a systematic and robust platform for qualitative analysis. These models facilitate deductive coding, enhance analytical rigor by mitigating common pitfalls, and have proven effective in core analytical tasks such as text clustering and summarization. Consequently, LLMs are utilized here to automate critical research processes, including the coding of primary data and the thematic modeling of the comprehensive dataset (Barros et al., 2025; Leca et al., 2024; Tai et al., 2024; Zenimoto et al., 2024).
Recent scholarship indicates that LLMs are becoming powerful tools in qualitative data analysis. Mathis et al. (2024) demonstrated that LLMs can efficiently generate robust themes from qualitative data, achieving a high degree of similarity to themes identified by human researchers. The validation of these models, along with formal evaluation methodologies, underscores their potential to enhance the rigor of qualitative research and their generalizability across disciplines. Echoing this, Leca et al. (2024) argued that emerging LLMs are primarily being applied to tasks such as coding, thematic analysis, and data classification, which yields significant advantages in efficiency and support for novice researchers. Nevertheless, the literature maintains that human expertise remains indispensable for data interpretation. Van Veggel et al. (2025) affirm this, positing that while artificial intelligence is a sophisticated analytical instrument, the development of theory and conceptual models remains firmly within the researcher’s domain, necessitating their unique theoretical sensitivity and expertise.
Recent advancements in LLMs like ChatGPT are poised to reshape the landscape of qualitative research, with applications evident at every stage of the process (Roberts et al., 2024). The role of these models is particularly salient in the domain of data analysis and the advancement of GT, as articulated by Sinha et al. (2024). They argue that the inherent structure of GT can serve as a guide for engineering specific and effective prompts, which consequently improves the precision of the LLM’s output. However, as Barros et al. (2025) caution, formidable challenges remain, notably the reliance on sophisticated prompt engineering, the potential for factual inaccuracies, and contextual limitations.
To address these limitations, the study’s methodological framework draws upon the seminal contributions of previous researchers. Specifically, it adapts the directives and task-specific prompts proposed by Sammon et al. (2024), Van Veggel et al. (2025), and Zhou et al. (2024) for guiding AI models through the distinct coding stages of GT.
Zhou et al. (2024) employed prompt engineering to structure a methodical approach for ChatGPT to perform coding consistent with GT principles. This was achieved by designing prompts encompassing five core elements: • “role setting, • task instruction, • context description, • input data, • and output indicator”.
This study selected three state-of-the-art AI models: ChatGPT, DeepSeek, and Gemini. These models are among the most influential LLMs to have been developed in recent years, originating from prominent technology firms in both the United States and China. Furthermore, they are accessible through both complimentary and commercial subscription tiers. The specific versions of these models utilized for the purposes of this research are detailed as follows: • ChatGPT 4o • Gemini Advanced Pro 2.5 • DeepSeek (Deepthink R1)
5.3. Coding Reliability
To ensure the trustworthiness of the coding schema, temporal stability was tested via a re-coding protocol conducted six months post-initialization (May vs. November 2025; see Appendices 4 and 6). Cross-verification focused on the persistence of subcategories, allowing for semantic flexibility (e.g., linguistic reformulation or category consolidation) as delineated in Appendix 7. Given that open coding subcategories provide the structural basis for axial and selective coding, confirming their presence across a six-month interval serves as a robust indicator of methodological reliability.
Results indicated high stability for the ChatGPT model, which reproduced all 13 initial subcategories (100%) and generated one new code (“Community Journalism”). The Gemini model also demonstrated strong reliability, replicating 91.7% (11/12) of initial codes while generating four new thematic areas in the second phase (e.g., “Gamification & Novel Interventions” and “eHealth & Medical Literacy”).
In contrast, the DeepSeek model showed significantly lower stability, with only 9 of the original 17 subcategories (53%) appearing in the re-coding phase. This reduced reliability is likely attributable to the model’s limitations in processing the full volume of analytical material. Therefore, the study restricted the final axial and selective coding phases to the ChatGPT and Gemini models only.
6. Results
6.1. Open Coding
The initial phase of analysis involved an open coding procedure applied to a corpus of 7,670 abstracts. This dataset was segmented into eight files for processing; seven of these files each contained 1,000 abstracts, while the eighth file consisted of the remaining 670.
A consistent prompt structure was systematically employed across the three LLMs. The design of this prompt was methodologically grounded in GT, ensuring the inclusion of five critical elements essential for this stage of coding: role setting, task instruction, context description, input data, and output indicator (see Appendices 1, 2, and 3).
Media Literacy and Expanding the Public Sphere: Main and Sub-categories According to the Three Models
An analysis of the data presented in Table 2, which is a synthesis of information from Tables 2, 3, and 4 (see Appendix 4), reveals the number of abstracts that each of the three models identified as pertinent to media literacy within the framework of an expanding public sphere. The proportion of these identified abstracts constitutes a range of 8.5% to 19.4% of the total abstracts. This quantitative variance suggests a notable divergence in the coding methodologies employed by the three models.
A significant congruence is observed between the approaches of ChatGPT and Gemini. In contrast, the DeepSeek model demonstrated a more constrained coding capacity, analyzing only 1,000 of the 7,670 total abstracts, from which it identified 194 as relevant. Conversely, ChatGPT identified 652 relevant abstracts, while Gemini identified a range of 750 to 850. The three models categorized the analyzed abstracts into five to seven main categories. These were further delineated into 12 to 17 subcategories, each encapsulating the core themes presented in the abstracts concerning the expansion of the public sphere.
6.2. Axial Coding
Building upon the results of the open coding phase, the subsequent axial coding stage utilized a specific prompt formula across the three models (see Appendices 1, 2, and 3).
The axial coding methodology used by ChatGPT, as depicted in Screenshot 1 (see Appendix 5), articulates the relationships between primary and subsidiary themes. An analysis of the main category, “media and information literacy,” indicates that it is informed by a diverse range of literacy skills. These skills are not only integral to effective interaction with media and information but are also foundational to broadening the public sphere, fostering civic participation, and advancing educational objectives. The axial coding for all remaining principal categories is detailed in Appendix 1.
Axial coding, performed with the assistance of Gemini (see Screenshot 2 - Appendix 5), produced the core category termed “The Empowerment Nexus.” This category conceptualizes the dynamic interplay between media literacy and the fostering of a more expansive and participatory public sphere. It posits that enhanced capabilities in critically evaluating and creating media empower individuals and communities for greater civic engagement, contribution to public discourse, and the ability to hold institutions accountable. The dimensions of this relationship are further articulated through several subcategories, which correlate specific media literacy skills with corresponding aspects of an expanded public sphere (Appendix 2). These are: • Foundational Media Literacy Skills for Public Engagement, • Literacy-Driven Expansion of Civic & Political Participation, • Broadening Access, Voice, and Representation through Media Literacy, • Nurturing Informed Communities and Discourse through Media Literacy.
The DeepSeek model was unable to execute the axial and selective coding phases due to recurrent server-busy errors (see Screenshot 3, Appendix 3 and 5). Consequently, the model’s application in this study was limited to the open coding stage.
6.3. Selective Coding
Building upon the results of the axial coding phase, the subsequent selective coding stage utilized a specific prompt formula (see Appendices 1, 2, and 3).
Screenshot 4 (see Appendix 5) offers a visual representation of the selective coding process, a key component of GT, as applied by ChatGPT. This process culminates in the development of a conceptual model titled “Empowered Media Citizenship.” The model elucidates the pathways through which educational inputs and prevailing contextual factors collaboratively foster civic empowerment. The conceptual framework is articulated through three core components, which directly address the analytical prompt: Process, Conditions, and Outcomes.
Following this breakdown, the output from ChatGPT elucidates how such a visual model is constructed, presenting it as the result of the axial and selective coding procedures. Building upon the selective coding phase, a theoretical model was developed to delineate the relationships between the pivotal category and other emergent categories, including the conditions, processes, and outcomes. The development of this model was explicitly constrained to the data and coding derived from the study’s sample abstracts, precluding the use of external sources through a specific prompt (see Appendices 1, 2, and 3).
Figure 1 illustrates the model generated by ChatGPT, which synthesizes the outcomes of the three-stage coding process applied to the research abstracts. Flowchart of media literacy and civic empowerment - ChatGPT. https://chatgpt.com/s/m_6824ad10145c8191a5858f6540407634
Appendix 2 presents the outcomes of the selective coding phase, conducted on 7,670 article abstracts using Gemini model. The analysis identified a core category defined as, “Civic Resilience in the Digital Age: Media Literacy as the Locus of an Empowered Public Sphere, Active Citizenship, and Trust Cultivation”. This category is conceptualized as a dynamic process wherein media literacy skills enhance an individual’s capacity to navigate complex information environments, critically evaluate sources, identify misinformation, and cultivate informed trust. These competencies are posited as prerequisites for both the preservation and expansion of a healthy public sphere.
Following the selective coding, and adhering to a prompt format also utilized with ChatGPT, Gemini was instructed to construct a model illustrating the relationships between the core category and other emergent categories, including their processes, conditions, and outcomes. Due to its inability to generate graphical outputs, Gemini provided a detailed textual description of a conceptual model titled, “The Engine of Civic Resilience: A Grounded Model for Media Literacy, Public Sphere Dynamics, and Empowered Citizenship”. Subsequently, A comprehensive code was obtained from Gemini to generate the diagram illustrated in Appendix 2. The code generated by Gemini (see Appendix 1) was exported to ChatGPT to produce an illustrative figure summarizing the Gemini coding operations, as depicted in Figure 2. (Gemini) The civic resilience engine: A grounded model of media literacy, public sphere dynamics, and empowered citizenship. https://chatgpt.com/s/m_687189bd1dac8191af648ddfda09df20
7. Discussion and Conclusion
This study employed open, axial, and selective coding on 7670 research abstracts using the (LLMs) ChatGPT, DeepSeek, and Gemini. While the coding frameworks of the models varied, a notable convergence was observed in identifying the abstracts focused on media literacy’s role in broadening the public sphere. The qualitative analysis yielded five to seven main thematic categories and 12 to 17 subcategories that encapsulated the conceptualizations of public sphere expansion discussed in the abstracts.
Extant literature (e.g., Bennis & Mouwafaq, 2025; Mizumoto & Teng, 2025; Shahzad & Iqbal, 2025) indicates that despite variations in their accuracy, coding quality, and efficiency, the three models have proven highly effective and precise within qualitative research methodologies. It is recommended that future directions focus on maintaining analytical rigor through a synergistic approach that combines AI capabilities with human scholarly judgment. This highlights the critical importance of meticulous oversight by expert researchers when incorporating Generative AI as a supportive instrument in the analysis of qualitative data.
An analysis of the categories generated during the open and axial coding phases indicates that the primary and secondary themes proposed by all three models—ChatGPT, DeepSeek, and Gemini—are conceptually aligned with the domain of MIL. The coding from ChatGPT most prominently featured categories concerning civic engagement and youth/student empowerment, digital literacy, and information and health literacy. The coding from DeepSeek primarily emphasized the integration of media literacy into education curricula, critical thinking, civic engagement, social media platforms, and public support. In Gemini’s output, the predominant categories included the role of social media platforms in public debate and opinion formation, the expansion of access to news and information via alternative media and open initiatives, and the critical consumption and production of communication messages.
These categories, as illustrated by the coding, are the focus of a significant body of literature analyzed under the present coding framework. As the influence of digital environments on youth participation in civic life is increasing, civic learning is shaped by navigating online information. Digital media literacy combined with civic education helps adolescents effectively participate in civic life (Bowyer & Kahne, 2020). Youth programs emphasize “youth voice” and structured participation to cultivate empowerment, belonging, and agency. Teen activism is framed as participatory politics, which uses online and offline means to mobilize and build civic identities (Jimenez et al., 2024; McDaniel, 2024). Digital literacy predicts young people’s involvement and civic participation in networked publics (Moon & Bai, 2024).
A layered competence model is needed, extending digital literacy beyond functional skills to include evaluative judgment. Algorithmic literacy, for example, focuses on interpreting systems that shape information exposure (Oeldorf-Hirsch & Neubaum, 2025). Higher education must support digital competence by integrating technical, cognitive, and socio-ethical skills (Anthonysamy & Sivakumar, 2023). Literacy is operationalized through professional development (Korona & Hutchison, 2023), curriculum alignment (Dezuanni, 2021), and pedagogies that cultivate critical consumption via reflective assignments (Estables & Vera, 2025; Fernández et al., 2022). Open designs, like MOOCs, also expand information literacy instruction across disciplines (Bussell & Larson, 2024; Moser et al., 2021).
Literacy is consequential for public health and debate, as platforms act as infrastructures shaping credibility. Health and media literacy influence adolescents’ real-world decision-making (Afshar et al., 2022). “Infodemic” frameworks stress evaluating trust to navigate misinformation (Sundelson et al., 2023). Platform dynamics, such as opinion leaders, shape information circulation and persuasion (Chew et al., 2023; Martin & MacDonald, 2020). Research on social platforms shows that mis/disinformation can intensify polarization, emphasizing the stakes of civic-literate engagement (Hassani et al., 2025; Rando-Cueto et al., 2025; Russo et al., 2025). Alternative initiatives broaden news access, particularly for underserved communities (Ebubedike et al., 2022; Meléndez-Labrador, 2023).
The ongoing scholarly discourse surrounding the main and subcategories of the public sphere has been significantly shaped by the rise of the internet and social networks. A prevailing perspective, articulated by scholars such as Flew (2022) and Schlesinger (2020), posits that the contemporary digital landscape may signal the decline, or even the complete collapse, of the traditional public sphere. This viewpoint suggests a contraction of the public sphere and the emergence of a “post-public sphere”. The foundation of this argument rests on the observation that a substantial portion of online communication is now mediated by a limited number of large-scale corporate digital platforms. These platforms are not merely commercial enterprises; their business models are predicated on the monetization of user data. Through constant surveillance, these platforms gather vast amounts of information to generate predictive behavioral analyses, which are then sold to advertisers for targeted messaging.
However, this conceptualization of the public sphere’s demise is not without its critics. Flew (2022) himself presents a nuanced counterargument, highlighting several factors that challenge the narrative of a slow death of the public sphere in the age of digital platforms. These counterarguments can be categorized into three main areas. Firstly, regulatory and legal frameworks are evolving to address the power of these digital platforms. Secondly, the professional function of news continues to play a vital role in informing the public. Thirdly, the influential role of experts, particularly during crises such as the COVID-19 pandemic, through both traditional and digital media, demonstrates the enduring relevance of informed public discourse.
According to Erstad and Amdam (2013), the advancement of media literacy is intrinsically linked to broader societal developments, including economic growth, the cultivation of civic awareness, and the maturation of the political landscape—all of which are integral components of a robust public sphere. The discourse surrounding media literacy has consequently shifted from a narrow focus on the media consumption habits of children and adolescents to more expansive considerations of social inclusion and active participation in public life. In this context, community media may play a crucial role by amplifying the voices of ordinary citizens and marginalized communities, groups often excluded from mainstream media narratives. As Caton-Rosser (2006) affirmed, a media-literate citizenry is fundamental to a healthy and vibrant democracy. This perspective of community media finds a contemporary parallel in the purported role of modern social media platforms. While the underlying technologies and communication mechanisms of the pre- and post-internet eras differ significantly, the conceptual link to fostering a participatory public sphere remains.
In line with the conclusions of Fuchs (2020) and Iosifidis (2011), the establishment of an inclusive and authentic public sphere is significantly advanced by the contributions of public service media and critical alternative media. Consequently, the expansion of this sphere necessitates several indispensable components. These include the promotion of MIL, the encouragement of civic engagement, the assurance of equitable representation for diverse social groups, and the provision of unrestricted access to information (Benson, 2008; Byeon, 2022; Cano-Orón et al., 2024; Dahlgren, 2005; Enjolras & Steen-Johnsen, 2017; Erstad & Amdam, 2013; Nangi et al., 2024; Papacharissi, 2002).
During the final coding and model-building phase, ChatGPT and Gemini were utilized to generate a concise title for the proposed model, derived from the three preceding coding phases. They also produced a conceptual map illustrating the relationships between the core category, associated processes, intervening enabling and hindering conditions, and desired outcomes.
As delineated in the ChatGPT axial coding framework (Appendix 1), the interactional process is analyzed through its causal conditions, intervening variables, and emergent consequences, centered on the pivotal category of media literacy and skills development. This process serves as a mechanism for critical engagement; however, its efficacy is contingent upon the cultivation of trust in credible media as a primary enabler of active citizenship. Conversely, the presence of misinformation and fake news functions as a significant structural impediment to empowerment. The terminal outcomes of this trajectory are defined as the realization of civic empowerment and the qualitative expansion of the public sphere.
The Gemini axial coding model (Appendix 2) further elucidates this interactional process as a sequential trajectory. This progression begins with media literacy development and advances through critical engagement and the mitigation of misinformation, which in turn fosters trust and civic activism. This iterative cycle culminates in the expanding of the public sphere and the establishment of a more resilient civic culture.
The core category, “Civic Resilience in the Digital Age,” functions as a comprehensive theoretical abstraction that synthesizes the operational dynamics of the “Empowerment Nexus” into a broader narrative of societal stability. Within this conceptual architecture, the effective functioning of the Empowerment Nexus—characterized by widespread media literacy—is positioned as a fundamental prerequisite for the cultivation of a robust public sphere and the mitigation of pervasive digital misinformation.
This relationship posits that the iterative interactions within the nexus facilitate the emergence of “Empowered Citizenship,” an outcome that transcends the mere expansion of the public domain to foster a more profound state of civic resilience. Ultimately, while the Empowerment Nexus provides the necessary operational mechanisms and interactional details, the core category articulates the overarching teleological vision, framing these components as essential for maintaining systemic trust and active engagement within a transformed digital landscape.
The model generated by ChatGPT (Figure 1) identifies “media literacy education” as the central category for achieving “empowered media citizenship.” This primary objective is influenced by a set of hindering conditions, most notably the proliferation of misinformation, fake news, and bias. Conversely, the enabling conditions consist of fostering trust in highly credible sources and promoting civic engagement. The projected outcome is an expansion of the public sphere, manifested through a discourse that ensures fair representation of societal groups and increased civic participation.
The Gemini Model, illustrated in Figure 2, posits a framework for “Civic Resilience in the Digital Age” centered on the core category of media literacy. This model identifies media literacy as the essential locus for cultivating an empowered public sphere, fostering active citizenship, and building societal trust. The efficacy of this model is predicated upon a set of interconnected enabling conditions: namely, the comprehensive development of media and digital literacy competencies, technical proficiency in utilizing relevant tools, and a critical understanding of the ethical dimensions and socio-emotional impacts of digital engagement. The model’s core processes involve an empowering dynamic where the application of these literacies yields distinct outcomes. Specifically, activating media literacy for public sphere engagement leads to transformative effects such as increased access to diverse information, an enhanced quality and inclusivity of public discourse, and the emergence of new platforms for deliberation. Concurrently, the practice of sharing critically evaluated information, combined with a strong ethical awareness, directly fosters active citizenship, stimulates political participation, and promotes the representation of marginalized groups.
Jürgen Habermas’s theory of the public sphere posits that the rise of mass communication has led to its structural transformation and subsequent “depoliticization of the public sphere”. He contends that the increasing interpenetration of the public and private realms, combined with the mass populace’s entry into the political domain, has produced a paradoxical effect. While this has ostensibly broadened citizen participation in public life, Habermas argues that this expansion has been largely stimulated by manipulative methods deployed through mass media, thereby eroding the sphere’s capacity for rational-critical debate (Lubenow, 2012; Mellen, 2010).
The proliferation of media disinformation, information disorder, and what is commonly termed “fake news” has escalated to unparalleled levels in the contemporary era. MIL serves as a foundational mechanism for cultivating individual and collective resilience to these phenomena. It enhances the capacity of citizens to critically evaluate and counter disinformation, thereby strengthening public oversight of institutional conduct. Moreover, it fosters a more pluralistic and diverse sphere of social media influence. Ultimately, a well-developed MIL is instrumental in shaping a robust public sphere, one that is less susceptible to the influences of commercial propaganda, ideological polarization, fascist rhetoric, and populist politics (Benson, 2008; Bleyer-Simon et al., 2024; Kumar & Maurya, 2024; Mansoor, 2024; Nasiri et al., 2024; Tomšič, 2023).
Through a qualitative analysis of summaries from 7,670 studies, both the ChatGPT and Gemini models demonstrated a consensus in outlining overarching frameworks for the interrelationship between media literacy enhancement and the public sphere. These conceptualizations, identified as “civic resilience in the digital age” and “empowered media citizenship,” resonate with extant literature in media, communication, and politics that associates the field with principles of human rights, digital citizenship, and sustainable development initiatives (Caton-Rosser, 2006; Enjolras & Steen-Johnsen, 2017; Erstad & Amdam, 2013; Han, 2024; Jayachandran, 2018; Lubenow, 2012; Mansoor, 2024; Mellen, 2010).
Furthermore, the proposed model advances a robust theoretical framework that supersedes traditional discourse on the expansion of the public sphere. By explicitly delineating opportunities and challenges, baseline conditions and anticipated outcomes, as well as requisite stakeholder roles and systemic barriers, it functions as a highly adaptable paradigm. Unconstrained by national borders or specific situational contexts, it can be broadly adapted while accommodating cultural nuances.
Central to operationalizing this model is Media and Information Literacy (MIL), which necessitates a holistic approach transcending conventional formal and informal educational paradigms. Cultivating MIL demands a multi-sectoral strategy, encompassing foundational familial structures alongside diverse governmental and private institutions. Meaningful and expedient outcomes rely intrinsically upon systemic coordination, strategic networking, and the implementation of medium- and long-term policies. Consequently, this requires sustained institutional support from policymakers, intellectual leaders, and broader socio-economic, technological, and political entities.
This model posits MIL as an instrumental means to foster civic resilience and media citizenship, thereby catalyzing a substantive expansion of the public sphere. Crucially, this perspective remains consonant with extant literature that conceptualizes MIL as an end in itself—an objective achieved through varied socio-cultural and educational mechanisms (Mansoor, 2023). Prior research framing MIL as a terminal objective provides the critical theoretical scaffolding necessary for this model. Ultimately, achieving MIL as a foundational end allows it to subsequently function as the operational means to achieve a genuine, rather than superficial, expansion of the public sphere.
8. Limitations and Future Research
The interpretation of the results from this study should be contextualized by its methodological limitations. Notwithstanding the detailed coding of an extensive dataset of over 7000 abstracts and the application of capable AI models for qualitative analysis, factors such as the sampling frame, the specific coding procedures, and the phrasing of prompts submitted to the LLMs necessarily circumscribe the findings.
Several methodological limitations warrant consideration regarding the generalizability of these findings. Primarily, the sampling frame relied exclusively on the Web of Science; despite its extensive repository, this single-database approach, combined with the established publication timeframe and an inherent English-language bias, may restrict the broader scope of the results.
The restriction of coding to abstracts was employed to ensure the broadest possible representation of the scientific literature. As abstracts encapsulate core theoretical concepts and main results, they provide sufficient material for the open coding phase of GT, which serves as the foundation for subsequent axial and selective coding. However, it is acknowledged that relying exclusively on abstracts may constrain conceptual richness. Given that GT traditionally mandates deep interpretative immersion, AI-powered analysis risks overlooking subtle nuances. Consequently, future research could adopt a multi-stage analytical framework utilizing AI models. For instance, conducting open, axial and selective coding phases for discrete sets of literature, such as in tranches of one thousand studies, could facilitate the discovery of additional subcategories and novel relationships. Such a process would enable a more profound analysis, thereby contributing to a more comprehensive model for understanding how the public sphere undergoes authentic, rather than artificially constructed, expansion. Furthermore, the strategic formulation of prompts for LLMs is a critical determinant of success when applying these AI models to qualitative methodologies, such as GT. This approach can yield more profound coding results in domains like public sphere studies and MIL, particularly when prompt design co-evolves with latest versions of these models and undergoes rigorous evaluation to mitigate potential bias.
The thematic visualizations generated from ChatGPT and Gemini offer a graphic representation of the relationships that underpin the expansion of the public sphere. However, the complexity and nuance of these categorical relationships could be significantly enhanced if future research were to adopt a multi-stage coding GT methodology across LLMs to discover additional subcategories and novel relationships.
The findings of this study underscore the imperative to cultivate a public sphere that reflects authentic, unadulterated public opinion—a significant endeavor for advancing principles of partnership and human rights. Nevertheless, the foundational concern articulated by Habermas regarding the potential for an expanding public sphere to dilute its capacity for rational critique warrants more rigorous investigation now than ever.
Supplemental Material
Supplemental Material - Toward a Model of Media Literacy and Public Sphere Expansion: A Grounded Theory Approach
Supplemental Material for Toward a Model of Media Literacy and Public Sphere Expansion: A Grounded Theory Approach by Hasan M. H. Mansoor in International Journal of Qualitative Methods.
Supplemental Material
Supplemental Material - Toward a Model of Media Literacy and Public Sphere Expansion: A Grounded Theory Approach
Supplemental Material for Toward a Model of Media Literacy and Public Sphere Expansion: A Grounded Theory Approach by Hasan M. H. Mansoor in International Journal of Qualitative Methods.
Supplemental Material
Supplemental Material - Toward a Model of Media Literacy and Public Sphere Expansion: A Grounded Theory Approach
Supplemental Material for Toward a Model of Media Literacy and Public Sphere Expansion: A Grounded Theory Approach by Hasan M. H. Mansoor in International Journal of Qualitative Methods.
Supplemental Material
Supplemental Material - Toward a Model of Media Literacy and Public Sphere Expansion: A Grounded Theory Approach
Supplemental Material for Toward a Model of Media Literacy and Public Sphere Expansion: A Grounded Theory Approach by Hasan M. H. Mansoor in International Journal of Qualitative Methods.
Supplemental Material
Supplemental Material - Toward a Model of Media Literacy and Public Sphere Expansion: A Grounded Theory Approach
Supplemental Material for Toward a Model of Media Literacy and Public Sphere Expansion: A Grounded Theory Approach by Hasan M. H. Mansoor in International Journal of Qualitative Methods.
Supplemental Material
Supplemental Material - Toward a Model of Media Literacy and Public Sphere Expansion: A Grounded Theory Approach
Supplemental Material for Toward a Model of Media Literacy and Public Sphere Expansion: A Grounded Theory Approach by Hasan M. H. Mansoor in International Journal of Qualitative Methods.
Supplemental Material
Supplemental Material - Toward a Model of Media Literacy and Public Sphere Expansion: A Grounded Theory Approach
Supplemental Material for Toward a Model of Media Literacy and Public Sphere Expansion: A Grounded Theory Approach by Hasan M. H. Mansoor in International Journal of Qualitative Methods.
Footnotes
Acknowledgment
The author acknowledges the use of ChatGPT, DeepSeek, and Gemini to conduct open, axial, and selective coding as a part of the methodology. The prompts used are included within the appendices of the study. The output from these prompts was a basic part of the results that aim to achieve the research objective. While the author acknowledges the usage of AI, he maintains that he is the sole author of this article and takes full responsibility for the content therein, as outlined in COPE recommendations. The author expresses his deepest appreciation and thanks to the Deanship of Scientific Research at King Saud University, represented by the Ongoing Research Funding program (ORF-2026-1525), for funding this research and their technical support.
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Ongoing Research Funding program, (ORF-2026-1525), King Saud University, Riyadh, Saudi Arabia.
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Data Availability Statement
The author confirms that all data generated or analysed during this study are included in this article and its appendices.
Supplemental Material
Supplemental material for this article is available online.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
